Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!usc!wuarchive!emory!hubcap!fpst
From: reynolds@park.bu.edu (John Reynolds)
Newsgroups: comp.parallel
Subject: Re: parallel data structures
Message-ID: <1991May23.180211.9343@hubcap.clemson.edu>
Date: 23 May 91 16:03:06 GMT
References: <1991May23.120025.17495@hubcap.clemson.edu>
Sender: news@bu.edu
Organization: Boston University Center for Adaptive Systems
Lines: 7
Approved: parallel@hubcap.clemson.edu
In-Reply-To: iahmad@wotan.top.cis.syr.edu's message of 22 May 91 23:45:58 GMT



Of course, decision trees tesselate feature space by dividing regions
into subregions recursively, placing hyperplanes that are
perpendicular to the coordinate axis corresponding to the current
decision point.  Discriminant functions provide more general way of
tesselating the space, which could include search tree tesselation.

A good comparison is ID-3 with ARTMAP.